System for Performing Resource Management and Control and Task Scheduling in Grid Computing

In order to make full use of resources in distributed parallel computing system, and reduce general time cost for performing the algorithm, task scheduling and resource management model and low-layer resource supervision-control model are designed and implemented. The unpredictable system behavior is resolved, which is caused by dynamic change resources. The dynamic join-in of resources is allowed. A large expansibility is characteristic. The genetic algorithm is used as scheduling policy, which enables this system to attain global optimum task scheduling. The GMA-based LDAP protocol is adopted in the system. The static data and dynamic ones are stored in directory service to reduce the interaction between the client and the server. Since the data can be obtained in one access, a shared resource environment with uniform programming interface is supplied to user.